In cities, around 25% of traffic is from people that got where they're going and are literally just searching for parking. Binghamton is investigating putting cameras on lots to detect open parking spaces and feed that information to drivers. Unfortunately, systems to detect parking spots using computer vision are extremely expensive for cities and universities.
What it does
SmartPark takes in camera feeds and detects how much of a parking lot is available.
How we built it
We wanted to defeat the expensive and complicated solutions with simplicity. We use a web interface to create boundaries around parking spots and we whipped up a backend utilizing OpenCV to probabilistically decide if there is a car present.
Challenges we ran into
Computer vision and HTTP headers are tricky
Accomplishments that we're proud of
Our idea for a radically simpler metric of whether a car is present worked, and we can actually bring this to the world after the hackathon.
What's next for Smart Park
We want this to live on, and we're going to make it open source so that cities and universities everywhere can use this technology at a low cost.